adding more steps

biocorecrg · Jul 15, 2024 · c3ef1a9 · c3ef1a9
1 parent 6c4f44b
commit c3ef1a9
Show file tree

Hide file tree

Showing 5 changed files with 49 additions and 33 deletions.
diff --git a/BioNextflow b/BioNextflow
diff --git a/mop_dna/mop_dna.nf b/mop_dna/mop_dna.nf
@@ -30,11 +30,16 @@ comparison                              : ${params.comparison}
 ********** reference has to be the genome *************
 reference                               : ${params.reference}
 snpeff_db                               : ${params.snpeff_db}
+clinvar_db                              : ${params.clinvar_db}
+clinvar_db_idx                          : ${params.clinvar_db_idx}
 output                                  : ${params.output}
 pars_tools                              : ${params.pars_tools}
 
-************************* Flows *******************************
+************************* SNPs ********************************
 clairS                                  : ${params.clairS}
+clairS_TO                               : ${params.clairS_TO}
+clair3                                  : ${params.clair3}
+***************** Structural Variants *************************
 sniffles                                : ${params.sniffles}
 ***************************************************************
 filter                                  : ${params.filter}
@@ -66,11 +71,13 @@ include { checkRef } from "${local_modules}"
 progPars = getParameters(params.pars_tools)
 
 include { RUN as RUN_CLAIRS } from "${subworkflowsDir}/snp_calling/clairS" addParams(LABEL: 'big_mem_cpus', EXTRAPARS: progPars["clairS--clairS"], OUTPUT: outputClairS)
+include { RUN as RUN_CLAIRSTO } from "${subworkflowsDir}/snp_calling/clairS_TO" addParams(LABEL: 'big_mem_cpus', EXTRAPARS: progPars["clairS_TO--clairS_TO"], OUTPUT: outputClairS)
 include { CHECK_FASTA } from "${subworkflowsDir}/misc/misc.nf"
 include { FAIDX as SAMTOOLS_FAIDX } from "${subworkflowsDir}/misc/samtools" 
 include { RUN as RUN_SNIFFLES } from "${subworkflowsDir}/structural_variant/sniffles.nf" addParams(EXTRAPARS: progPars["sniffles--sniffles"], OUTPUT:outputSniffles, LABEL:'big_mem_cpus', CONTAINER:'quay.io/biocontainers/sniffles:2.3.3--pyhdfd78af_0')
-include { RUN_MULTI as RUN_MULTI_SNIFFLES }  from "${subworkflowsDir}/structural_variant/sniffles.nf" addParams(EXTRAPARS: progPars["sniffles_multi--sniffles"], OUTPUT:outputSniffles, LABEL:'big_mem_cpus', CONTAINER:'quay.io/biocontainers/sniffles:2.3.3--pyhdfd78af_0')
-include { ANN as SNPEFF_ANN }  from "${subworkflowsDir}/var_annotation/snpeff.nf" addParams(EXTRAPARS: progPars["commong--snpEff"], OUTPUT:outputAnnotation)
+//include { RUN_MULTI as RUN_MULTI_SNIFFLES }  from "${subworkflowsDir}/structural_variant/sniffles.nf" addParams(EXTRAPARS: progPars["sniffles_multi--sniffles"], OUTPUT:outputSniffles, LABEL:'big_mem_cpus', CONTAINER:'quay.io/biocontainers/sniffles:2.3.3--pyhdfd78af_0')
+include { ANN as SNPEFF_ANN }  from "${subworkflowsDir}/var_annotation/snpeff.nf" addParams(EXTRAPARS: progPars["common--snpEff"], OUTPUT:outputAnnotation)
+include { ANN as SNPSIFT_ANN } from "${subworkflowsDir}/var_annotation/snpsift.nf" addParams(OUTPUT:outputAnnotation)
 include { VIEW as BCFTOOLS_VIEW }  from "${subworkflowsDir}/snp_calling/bcftools.nf" addParams(EXTRAPARS: progPars["filter--bcftools-view"], OUTPUT:outputFiltered)
 include { REPORT as MULTIQC } from "${subworkflowsDir}/reporting/multiqc" addParams(OUTPUT:outputMultiQC)
 
@@ -128,20 +135,38 @@ workflow {
    		}
 	} 
 
+    if (params.clairS_TO != "NO") {
+                out_clairsto = RUN_CLAIRSTO(bams.join(bais), checked_genome, faidx)
+
+   	vcf_clairSTO = out_clairsto.snps.map{
+   		   [ "${it[0]}--clairS---snps", it[1][0] ]
+   		}.mix(out_clairsto.indels.map{
+           [ "${it[0]}--clairS---indels", it[1][0] ]
+        })
+	} 
     // Structural variants
     if (params.sniffles != "NO") {
-    	vcf_sniffles = sniffles_flow(bams, bais, comparisons, checked_genome).map{
+    	vcf_sniffles = sniffles_flow(bams, bais, checked_genome).map{
    			[ "${it[0]}--sniffles", it[1]]
    		}
     }
 
-    all_vcf = vcf_clairS.mix(vcf_sniffles)
+    all_vcf = vcf_clairS.mix(vcf_sniffles).mix(vcf_clairSTO)
+
 	if (params.filter != "NO") {
 		filt_vcf = BCFTOOLS_VIEW(all_vcf)
 	}
 
-    annos = SNPEFF_ANN(filt_vcf, params.snpeff_db, snpEff_data, snpEff_config).csv_summary.map{it[1]}
+    out_snpeff = SNPEFF_ANN(filt_vcf, params.snpeff_db, snpEff_data, snpEff_config)
+    annos = out_snpeff.csv_summary.map{it[1]}
 
+    if (params.clinvar_db != "" && params.clinvar_db_idx != "") {
+        clinvarDB = channel.fromPath(params.clinvar_db, checkIfExists: true)
+        clinvarDBi = channel.fromPath(params.clinvar_db_idx, checkIfExists: true)
+   	    SNPSIFT_ANN(out_snpeff.annotation, clinvarDB, clinvarDBi)
+    }
+
+
 	MULTIQC(annos.collect())
 }
 
@@ -179,31 +204,14 @@ workflow sniffles_flow {
     take:
     bam
     bais
-    comparisons
     reference
 
 
     main:
-    bam_comp = mapIDPairs(comparisons, bam).map{
-    	[ "${it[0]}--${it[1]}", it[2], it[3] ]  
-    }
-
-
-    bai_comp = mapIDPairs(comparisons, bais).map{
-    	[ "${it[0]}--${it[1]}", it[2], it[3] ]  
-    }
-
-    data = bam_comp.join(bai_comp)    
     out = RUN_SNIFFLES(bam, bais, reference)
-
-    snf_comp = mapIDPairs(comparisons, out.snf).map{
-    	[ "${it[0]}--${it[1]}", [ it[2], it[3] ] ]  
-    }
-
-    out = RUN_MULTI_SNIFFLES(snf_comp)
-
+
     emit:
-    	out
+    	out.vcf
 
 }
 

diff --git a/mop_dna/params.yaml b/mop_dna/params.yaml
@@ -1,18 +1,26 @@
 input_path: "${projectDir}/test_dataset/"
-comparison: "${projectDir}/comparison.tsv"
 
 ##reference: "${projectDir}/GRCh38.primary_assembly.genome.fa"
 reference: "/db/gencode/human/release_38/GRCh38.primary_assembly.genome.fa.gz"
 snpeff_db: "GRCh38.105"
+clinvar_db: "${projectDir}/clinvar/clinvar_20240630.vcf.gz"
+clinvar_db_idx: "${projectDir}/clinvar/clinvar_20240630.vcf.gz.tbi"
 output: "${projectDir}/output_dna"
 pars_tools: "${projectDir}/tools_opt.tsv"
 
-# flows
-clairS: "YES"
+# clair flows
+## clairS for comparing tumor/normal samples
+clairS: "NO"
+comparison: "${projectDir}/comparison.tsv"
+## clairS-TO (tumor only)
+clairS_TO: "YES"
+
+## sniffles for structural variants
 sniffles: "YES"
+
+## Commmon flows
 filter: "YES"
 annotation: "YES"
 
-# epinano plots
 hook: ""
 email: ""
diff --git a/mop_dna/tools_opt.tsv b/mop_dna/tools_opt.tsv
@@ -1,6 +1,6 @@
 #flows	tool	extrapars
 sniffles	sniffles	"--tandem-repeats ${projectDir}/anno/human_GRCh38_no_alt_analysis_set.trf.bed"
-sniffles_multi	sniffles	""
-clairS	clairS	"-p ont_r9_guppy"
+clairS	clairS	""
+clairS_TO	clairS_TO	"-p ont_r10_dorado_sup_5khz"
 filter	bcftools-view	"-f PASS"
 common	snpEff	""	
diff --git a/outdirs.nf b/outdirs.nf
@@ -18,7 +18,7 @@ outputTomboFlow      = "${params.output}/tombo_flow"
 outputModPhredFlow   = "${params.output}/modphred_flow"
 
 // mop_dna
-outputClairS         = "${params.output}/clairS_flow"
+outputClairS         = "${params.output}/clair_flow"
 outputSniffles       = "${params.output}/sniffles_flow"
 outputAnnotation     = "${params.output}/annotation"
 outputFiltered       = "${params.output}/filtered_vcf"
+2 −2		subworkflows/snp_calling/clairS.nf
+85 −0		subworkflows/snp_calling/clairS_TO.nf
+1 −1		subworkflows/var_annotation/snpeff.nf
+72 −0		subworkflows/var_annotation/snpsift.nf